BigP3BCI Study M is a derivative EEG dataset containing P300-based brain-computer interface recordings from 21 ALS subjects using a 9x8 adaptive/checkerboard visual stimulus paradigm. This dataset is part of the BigP3BCI project, the largest public P300 BCI dataset, and includes preprocessed EEG data acquired at 256 Hz from 16 channels using standard 10-20 electrode montage. The data supports P300 event-related potential detection and speller BCI applications with target and non-target stimulus classifications.